Genomic data counts study-view endpoint (#10300)

* implement getting CNA types for a gene specific

* add tests for genomic-data-counts endpoint

* change CopyNumberDataCounterFilter to StudyViewFilter and GenomicDataCountFilter

* change to molecularDataService to fetch data

* add filter logic for DISCRETE molecular profiles

* Refactor

* count NA and clean up codes

* Updates

* Update StudyViewServiceImplTest.java

---------

Co-authored-by: Qi-Xuan Lu <qlu@carisls.com>
Co-authored-by: Karthik <kalletlakarthik@gmail.com>

model/src/main/java/org/cbioportal/model/GenomicDataCountItem.java

@@ -0,0 +1,35 @@
+package org.cbioportal.model;
+
+import java.io.Serializable;
+import java.util.List;
+
+public class GenomicDataCountItem implements Serializable {
+
+    private String hugoGeneSymbol;
+    private String profileType;
+    private List<GenomicDataCount> counts;
+
+    public String getHugoGeneSymbol() {
+        return hugoGeneSymbol;
+    }
+
+    public void setHugoGeneSymbol(String hugoGeneSymbol) {
+        this.hugoGeneSymbol = hugoGeneSymbol;
+    }
+
+    public String getProfileType() {
+        return profileType;
+    }
+
+    public void setProfileType(String profileType) {
+        this.profileType = profileType;
+    }
+
+    public List<GenomicDataCount> getCounts() {
+        return counts;
+    }
+
+    public void setCounts(List<GenomicDataCount> counts) {
+        this.counts = counts;
+    }
+}

service/src/main/java/org/cbioportal/service/StudyViewService.java

@@ -1,6 +1,8 @@
 package org.cbioportal.service;
 
+import org.apache.commons.math3.util.Pair;
 import org.cbioportal.model.*;
+import org.cbioportal.service.exception.MolecularProfileNotFoundException;
 import org.cbioportal.service.exception.StudyNotFoundException;
 
 import java.util.List;
@@ -19,6 +21,8 @@ List<AlterationCountByGene> getStructuralVariantAlterationCountByGenes(List<Stri
     List<CopyNumberCountByGene> getCNAAlterationCountByGenes(List<String> studyIds, List<String> sampleIds, AlterationFilter annotationFilter)
         throws StudyNotFoundException;
 
+    List<GenomicDataCountItem> getCNAAlterationCountsByGeneSpecific(List<String> studyIds, List<String> sampleIds, List<Pair<String, String>> genomicDataFilters);
+
     List<GenericAssayDataCountItem> fetchGenericAssayDataCounts(List<String> sampleIds, List<String> studyIds, List<String> stableIds, List<String> profileTypes);
 
 }

service/src/main/java/org/cbioportal/service/impl/StudyViewServiceImpl.java

@@ -2,6 +2,7 @@
 
 import org.apache.commons.collections4.map.MultiKeyMap;
 import org.apache.commons.lang3.StringUtils;
+import org.apache.commons.math3.util.Pair;
 import org.cbioportal.model.*;
 import org.cbioportal.model.util.Select;
 import org.cbioportal.persistence.AlterationRepository;
@@ -15,6 +16,7 @@
 import java.util.*;
 import java.util.function.Function;
 import java.util.stream.Collectors;
+import java.util.stream.Stream;
 
 @Service
 public class StudyViewServiceImpl implements StudyViewService {
@@ -37,6 +39,12 @@ public class StudyViewServiceImpl implements StudyViewService {
     @Autowired
     private AlterationRepository alterationRepository;
 
+    @Autowired
+    private GeneService geneService;
+
+    @Autowired 
+    private MolecularDataService molecularDataService;
+
     @Override
     public List<GenomicDataCount> getGenomicDataCounts(List<String> studyIds, List<String> sampleIds) {
         List<MolecularProfileCaseIdentifier> molecularProfileSampleIdentifiers =
@@ -154,7 +162,7 @@ public List<CopyNumberCountByGene> getCNAAlterationCountByGenes(List<String> stu
         throws StudyNotFoundException {
         List<MolecularProfileCaseIdentifier> caseIdentifiers =
             molecularProfileService.getFirstDiscreteCNAProfileCaseIdentifiers(studyIds, sampleIds);
-        Select<CNA> cnaTypes = Select.byValues(CNA_TYPES_AMP_AND_HOMDEL);
+
         List<CopyNumberCountByGene> copyNumberCountByGenes = alterationCountService.getSampleCnaGeneCounts(
             caseIdentifiers,
             Select.all(),
@@ -186,6 +194,96 @@ public List<CopyNumberCountByGene> getCNAAlterationCountByGenes(List<String> stu
         return copyNumberCountByGenes;
     }
 
+    @Override
+    public List<GenomicDataCountItem> getCNAAlterationCountsByGeneSpecific(List<String> studyIds, 
+                                                                           List<String> sampleIds,
+                                                                           List<Pair<String, String>> genomicDataFilters) {
+
+        List<MolecularProfile> molecularProfiles = molecularProfileService.getMolecularProfilesInStudies(studyIds,
+            "SUMMARY");
+
+        Map<String, List<MolecularProfile>> molecularProfileMap = molecularProfileUtil
+            .categorizeMolecularProfilesByStableIdSuffixes(molecularProfiles);
+
+        Set<String> hugoGeneSymbols = genomicDataFilters.stream().map(Pair::getKey)
+            .collect(Collectors.toSet());
+
+        Map<String, Integer> geneSymbolIdMap = geneService
+            .fetchGenes(new ArrayList<>(hugoGeneSymbols), "HUGO_GENE_SYMBOL",
+                "SUMMARY")
+            .stream().collect(Collectors.toMap(Gene::getHugoGeneSymbol, Gene::getEntrezGeneId));
+
+        return genomicDataFilters
+            .stream()
+            .flatMap(gdFilter -> {
+                GenomicDataCountItem genomicDataCountItem = new GenomicDataCountItem();
+                String hugoGeneSymbol = gdFilter.getKey();
+                String profileType = gdFilter.getValue();
+                genomicDataCountItem.setHugoGeneSymbol(hugoGeneSymbol);
+                genomicDataCountItem.setProfileType(profileType);
+
+                List<String> stableIds = Arrays.asList(geneSymbolIdMap.get(hugoGeneSymbol).toString());
+
+                Map<String, String> studyIdToMolecularProfileIdMap = molecularProfileMap
+                    .getOrDefault(profileType, new ArrayList<MolecularProfile>()).stream()
+                    .collect(Collectors.toMap(MolecularProfile::getCancerStudyIdentifier,
+                        MolecularProfile::getStableId));
+
+                List<String> mappedSampleIds = new ArrayList<>();
+                List<String> mappedProfileIds = new ArrayList<>();
+
+                for (int i = 0; i < sampleIds.size(); i++) {
+                    String studyId = studyIds.get(i);
+                    if (studyIdToMolecularProfileIdMap.containsKey(studyId)) {
+                        mappedSampleIds.add(sampleIds.get(i));
+                        mappedProfileIds.add(studyIdToMolecularProfileIdMap.get(studyId));
+                    }
+                }
+
+                if (mappedSampleIds.isEmpty()) {
+                    return Stream.of();
+                }
+
+                List<GeneMolecularData> geneMolecularDataList = molecularDataService.getMolecularDataInMultipleMolecularProfiles(mappedProfileIds, mappedSampleIds,
+                        stableIds.stream().map(Integer::parseInt).collect(Collectors.toList()), "SUMMARY");
+
+                List<GenomicDataCount> genomicDataCounts = geneMolecularDataList
+                    .stream()
+                    .filter(g -> StringUtils.isNotEmpty(g.getValue()) && !g.getValue().equals("NA"))
+                    .collect(Collectors.groupingBy(GeneMolecularData::getValue))
+                    .entrySet()
+                    .stream()
+                    .map(entry -> {
+                        Integer alteration = Integer.valueOf(entry.getKey());
+                        List<GeneMolecularData> geneMolecularData = entry.getValue();
+                        int count = geneMolecularData.size();
+
+                        String label = CNA.getByCode(alteration.shortValue()).getDescription();
+
+                        GenomicDataCount genomicDataCount = new GenomicDataCount();
+                        genomicDataCount.setLabel(label);
+                        genomicDataCount.setValue(String.valueOf(alteration));
+                        genomicDataCount.setCount(count);
+
+                        return genomicDataCount;
+                    }).collect(Collectors.toList());
+
+                int totalCount = genomicDataCounts.stream().mapToInt(GenomicDataCount::getCount).sum();
+                int naCount = sampleIds.size() - totalCount;
+
+                if (naCount > 0) {
+                    GenomicDataCount genomicDataCount = new GenomicDataCount();
+                    genomicDataCount.setLabel("NA");
+                    genomicDataCount.setValue("NA");
+                    genomicDataCount.setCount(naCount);
+                    genomicDataCounts.add(genomicDataCount);
+                }
+
+                genomicDataCountItem.setCounts(genomicDataCounts);
+                return Stream.of(genomicDataCountItem);
+            }).collect(Collectors.toList());
+    };
+
     @Override
     public List<GenericAssayDataCountItem> fetchGenericAssayDataCounts(List<String> sampleIds, List<String> studyIds,
                                                                List<String> stableIds, List<String> profileTypes) {
@@ -261,5 +359,4 @@ public List<GenericAssayDataCountItem> fetchGenericAssayDataCounts(List<String>
             })
             .collect(Collectors.toList());
     }
-
 }

service/src/test/java/org/cbioportal/service/impl/BaseServiceImplTest.java

@@ -59,5 +59,7 @@ public class BaseServiceImplTest {
     public static final String CATEGORY_VALUE_1 = "category_value1";
     public static final String CATEGORY_VALUE_2 = "category_value2";
     public static final String EMPTY_VALUE_1 = "";
-    public static final String EMPTY_VALUE_2 = "NA";    
+    public static final String EMPTY_VALUE_2 = "NA";
+    public static final String PROFILE_TYPE_1 = "profile_type1";
+    public static final String PROFILE_TYPE_2 = "profile_type2";
 }

service/src/test/java/org/cbioportal/service/impl/StudyViewServiceImplTest.java

@@ -14,10 +14,7 @@
 import org.mockito.Spy;
 import org.mockito.junit.MockitoJUnitRunner;
 
-import java.util.ArrayList;
-import java.util.Arrays;
-import java.util.Collections;
-import java.util.List;
+import java.util.*;
 
 import static org.mockito.ArgumentMatchers.*;
 
@@ -41,6 +38,10 @@ public class StudyViewServiceImplTest extends BaseServiceImplTest {
     private SignificantCopyNumberRegionService significantCopyNumberRegionService;
     @Mock
     private GenericAssayService genericAssayService;
+    @Mock
+    private MolecularDataService molecularDataService;
+    @Mock
+    private GeneService geneService;
     private AlterationFilter alterationFilter = new AlterationFilter();
 
     @Test
@@ -255,6 +256,79 @@ public void getCNAAlterationCountByGenes() throws Exception {
         Assert.assertEquals(1, result.size());
     }
 
+    @Test
+    public void getCNAAlterationCountByGeneSpecific() throws Exception {
+
+        List<String> sampleIds = Arrays.asList(BaseServiceImplTest.SAMPLE_ID1, BaseServiceImplTest.SAMPLE_ID2, BaseServiceImplTest.SAMPLE_ID3);
+        List<String> studyIds = Collections.nCopies(3, BaseServiceImplTest.STUDY_ID);
+        List<Pair<String, String>> genomicDataFilters = new ArrayList<>();
+        Pair<String, String> genomicDataFilter1 = new Pair<>(BaseServiceImplTest.HUGO_GENE_SYMBOL_1, BaseServiceImplTest.PROFILE_TYPE_1);
+        Pair<String, String> genomicDataFilter2 = new Pair<>(BaseServiceImplTest.HUGO_GENE_SYMBOL_2, BaseServiceImplTest.PROFILE_TYPE_2);
+        genomicDataFilters.add(genomicDataFilter1);
+        genomicDataFilters.add(genomicDataFilter2);
+
+        List<MolecularProfile> molecularProfiles = new ArrayList<>();
+        MolecularProfile molecularProfile = new MolecularProfile();
+        molecularProfile.setCancerStudyIdentifier(BaseServiceImplTest.STUDY_ID);
+        molecularProfile.setStableId(BaseServiceImplTest.STUDY_ID + "_" + BaseServiceImplTest.MOLECULAR_PROFILE_ID_A);
+        molecularProfiles.add(molecularProfile);
+
+        Mockito.when(molecularProfileService.getMolecularProfilesInStudies(studyIds, "SUMMARY"))
+            .thenReturn(molecularProfiles);
+
+        Map<String, List<MolecularProfile>> molecularProfileMap = new HashMap<>();
+        molecularProfileMap.put(BaseServiceImplTest.PROFILE_TYPE_1, molecularProfiles);
+        molecularProfileMap.put(BaseServiceImplTest.PROFILE_TYPE_2, molecularProfiles);
+
+        Mockito.when(molecularProfileUtil.categorizeMolecularProfilesByStableIdSuffixes(molecularProfiles))
+            .thenReturn(molecularProfileMap);
+
+        List<Gene> genes = new ArrayList<>();
+        Gene gene1 = new Gene();
+        gene1.setEntrezGeneId(BaseServiceImplTest.ENTREZ_GENE_ID_1);
+        gene1.setHugoGeneSymbol(BaseServiceImplTest.HUGO_GENE_SYMBOL_1);
+        Gene gene2 = new Gene();
+        gene2.setEntrezGeneId(BaseServiceImplTest.ENTREZ_GENE_ID_2);
+        gene2.setHugoGeneSymbol(BaseServiceImplTest.HUGO_GENE_SYMBOL_2);
+        genes.add(gene1);
+        genes.add(gene2);
+
+        Mockito.when(geneService.fetchGenes(anyList(), anyString(), anyString()))
+                .thenReturn(genes);
+
+        List<GeneMolecularData> geneMolecularData = new ArrayList<>();
+        GeneMolecularData geneMolecularData1 = new GeneMolecularData();
+        geneMolecularData1.setValue("-2");
+        GeneMolecularData geneMolecularData2 = new GeneMolecularData();
+        geneMolecularData2.setValue("-2");
+        GeneMolecularData geneMolecularData3 = new GeneMolecularData();
+        geneMolecularData3.setValue("2");
+        GeneMolecularData geneMolecularData4 = new GeneMolecularData();
+        geneMolecularData4.setValue("2");
+        geneMolecularData.add(geneMolecularData1);
+        geneMolecularData.add(geneMolecularData2);
+        geneMolecularData.add(geneMolecularData3);
+        geneMolecularData.add(geneMolecularData4);
+
+        Mockito.when(molecularDataService.getMolecularDataInMultipleMolecularProfiles(
+            anyList(),
+            anyList(),
+            anyList(),
+            anyString()))
+            .thenReturn(geneMolecularData);
+
+        List<GenomicDataCountItem> result = studyViewService.getCNAAlterationCountsByGeneSpecific(studyIds, sampleIds, genomicDataFilters);
+        Assert.assertEquals(2, result.size());
+        Assert.assertEquals(BaseServiceImplTest.HUGO_GENE_SYMBOL_1, result.get(0).getHugoGeneSymbol());
+        Assert.assertEquals(BaseServiceImplTest.PROFILE_TYPE_1, result.get(0).getProfileType());
+        Assert.assertEquals(2, result.get(0).getCounts().get(0).getCount().intValue());
+        Assert.assertEquals(2, result.get(0).getCounts().get(1).getCount().intValue());
+        Assert.assertEquals(BaseServiceImplTest.HUGO_GENE_SYMBOL_2, result.get(1).getHugoGeneSymbol());
+        Assert.assertEquals(BaseServiceImplTest.PROFILE_TYPE_2, result.get(1).getProfileType());
+        Assert.assertEquals(2, result.get(1).getCounts().get(0).getCount().intValue());
+        Assert.assertEquals(2, result.get(1).getCounts().get(1).getCount().intValue());
+    }
+
     @Test
     public void fetchGenericAssayDataCounts() throws Exception {
 

web/src/main/java/org/cbioportal/web/StudyViewController.java

@@ -9,14 +9,14 @@
 import org.apache.commons.lang3.tuple.ImmutablePair;
 import org.apache.commons.math3.stat.correlation.PearsonsCorrelation;
 import org.apache.commons.math3.stat.correlation.SpearmansCorrelation;
+import org.apache.commons.math3.util.Pair;
 import org.cbioportal.model.*;
 import org.cbioportal.service.*;
 import org.cbioportal.service.exception.StudyNotFoundException;
 import org.cbioportal.service.util.ClinicalAttributeUtil;
 import org.cbioportal.web.config.annotation.InternalApi;
 import org.cbioportal.model.AlterationFilter;
 import org.cbioportal.web.parameter.*;
-import org.cbioportal.web.parameter.sort.ClinicalDataSortBy;
 import org.cbioportal.web.util.*;
 import org.springframework.beans.factory.annotation.Autowired;
 import org.springframework.cache.annotation.Cacheable;
@@ -861,6 +861,47 @@ public ResponseEntity<List<GenomicDataBin>> fetchGenomicDataBinCounts(
         return new ResponseEntity<>(studyViewFilterApplier.getDataBins(dataBinMethod, interceptedGenomicDataBinCountFilter), HttpStatus.OK);
     }
 
+    @PreAuthorize("hasPermission(#involvedCancerStudies, 'Collection<CancerStudyId>', T(org.cbioportal.utils.security.AccessLevel).READ)")
+    @RequestMapping(value = "/genomic-data-counts/fetch", method = RequestMethod.POST,
+        consumes = MediaType.APPLICATION_JSON_VALUE, produces = MediaType.APPLICATION_JSON_VALUE)
+    @ApiOperation("Fetch genomic data counts by GenomicDataCountFilter")
+    public ResponseEntity<List<GenomicDataCountItem>> fetchGenomicDataCounts(
+        @ApiParam(required = true, value = "Genomic data count filter") @Valid @RequestBody(required = false) GenomicDataCountFilter genomicDataCountFilter,
+        @ApiIgnore // prevent reference to this attribute in the swagger-ui interface
+        @RequestAttribute(required = false, value = "involvedCancerStudies") Collection<String> involvedCancerStudies,
+        @ApiParam(required = true, value = "Intercepted Genomic Data Count Filter")
+        @ApiIgnore
+        @Valid @RequestAttribute(required = false, value = "interceptedGenomicDataCountFilter") GenomicDataCountFilter interceptedGenomicDataCountFilter
+    ) throws StudyNotFoundException {
+        List<GenomicDataFilter> gdFilters = interceptedGenomicDataCountFilter.getGenomicDataFilters();
+        StudyViewFilter studyViewFilter = interceptedGenomicDataCountFilter.getStudyViewFilter();
+        // when there is only one filter, it means study view is doing a single chart filter operation
+        // remove filter from studyViewFilter to return all data counts
+        // the reason we do this is to make sure after chart get filtered, user can still see unselected portion of the chart
+        if (gdFilters.size() == 1) {
+            studyViewFilterUtil.removeSelfFromGenomicDataFilter(
+                gdFilters.get(0).getHugoGeneSymbol(), 
+                gdFilters.get(0).getProfileType(), 
+                studyViewFilter);
+        }
+        List<SampleIdentifier> filteredSampleIdentifiers = studyViewFilterApplier.apply(studyViewFilter);
+
+        if (filteredSampleIdentifiers.isEmpty()) {
+            return new ResponseEntity<>(new ArrayList<>(), HttpStatus.OK);
+        }
+
+        List<String> studyIds = new ArrayList<>();
+        List<String> sampleIds = new ArrayList<>();
+        studyViewFilterUtil.extractStudyAndSampleIds(filteredSampleIdentifiers, studyIds, sampleIds);
+
+        List<GenomicDataCountItem> result = studyViewService.getCNAAlterationCountsByGeneSpecific(
+            studyIds,
+            sampleIds,
+            gdFilters.stream().map(gdFilter -> new Pair<>(gdFilter.getHugoGeneSymbol(), gdFilter.getProfileType())).collect(Collectors.toList()));
+
+        return new ResponseEntity<>(result, HttpStatus.OK);
+    }
+
     @PreAuthorize("hasPermission(#involvedCancerStudies, 'Collection<CancerStudyId>', T(org.cbioportal.utils.security.AccessLevel).READ)")
     @RequestMapping(value = "/generic-assay-data-counts/fetch", method = RequestMethod.POST, consumes = MediaType.APPLICATION_JSON_VALUE, produces = MediaType.APPLICATION_JSON_VALUE)
     @ApiOperation("Fetch generic assay data counts by study view filter")

web/src/main/java/org/cbioportal/web/parameter/GenomicDataCountFilter.java

@@ -0,0 +1,26 @@
+package org.cbioportal.web.parameter;
+
+import java.io.Serializable;
+import java.util.List;
+
+public class GenomicDataCountFilter implements Serializable {
+
+    private List<GenomicDataFilter> genomicDataFilters;
+    private StudyViewFilter studyViewFilter;
+
+    public List<GenomicDataFilter> getGenomicDataFilters() {
+        return genomicDataFilters;
+    }
+
+    public void setGenomicDataFilters(List<GenomicDataFilter> genomicDataFilters) {
+        this.genomicDataFilters = genomicDataFilters;
+    }
+
+    public StudyViewFilter getStudyViewFilter() {
+        return studyViewFilter;
+    }
+
+    public void setStudyViewFilter(StudyViewFilter studyViewFilter) {
+        this.studyViewFilter = studyViewFilter;
+    }
+}